A feature model based business process model query method

Abstract

The application discloses a business process model query method based on a feature model, and belongs to the technical field of business process model search, and the method comprises the following steps: classifying all business process models in a business process model library to obtain a feature model; optimizing the feature model to obtain an optimized feature model; obtaining an input model, and performing query according to the optimized feature model to obtain a query result. The method can improve the accuracy and efficiency of business process model query.

Description

TECHNICAL FIELD

[0001] The application relates to a feature model-based business process model query method and belongs to the technical field of business process model search. BACKGROUND

[0002] With the continuous increase and complication of business process models, the problem of low management efficiency of the process model library has become a concern of people, wherein, for the model library with a large number of models, a corresponding model matching method is often used to help people screen target models from the model library. However, the traditional model matching method is to give an input model in the model library and then perform real-time calculation for matching. Such a calculation process has the following problems:

[0003] On the one hand, real-time matching does not pre-process the models, so that all the models need to be traversed in the model matching process, resulting in low query efficiency.

[0004] On the other hand, in the case of non-accurate search in the determination of the search range, when the calculation of the traversal matching of the models is performed to a certain extent, it can be relatively accurately judged whether the model of this matching is the model to be queried. However, the traditional business process model similarity algorithm requires the traversal matching of nodes and edges, so that an early termination algorithm and condition setting are lacked, and in the case of a large number of models, the efficiency problem will also be caused.

[0005] The above two problems will become more acute contradictions with the increase of the number of models in the model library. In addition, the traditional business process model similarity technology will introduce a synonym library and a word net when measuring semantic similarity. The synonym library and the word net are usually general and involve less professional knowledge. The business process model is domain-specific and involves the correlation between professional terms, which is difficult to calculate by using the traditional semantic similarity measurement method. Therefore, a knowledge structure involving more relationships between professional terms and more professional knowledge is needed to further improve the accuracy of similarity measurement. At the same time, since the knowledge structure is a relationship form rather than a simple connection between words, it has stronger expansibility and connectivity and can solve the problem of different abstraction levels in model matching. SUMMARY

[0006] The application aims to provide a feature model-based business process model query method, which can improve the accuracy and efficiency of business process model query.

[0007] To achieve the above-mentioned purpose, the application provides the following technical scheme:

[0008] A feature model-based business process model query method comprises the following steps:

[0009] Classify all business process models in the business process model library to obtain a feature model;

[0010] Optimize the feature model to obtain an optimized feature model;

[0011] Obtain an input model and query according to the optimized feature model to obtain a query result.

[0012] Further, classifying all business process models in the business process model library to obtain a feature model comprises:

[0013] Obtain similar business process models in the business process model library;

[0014] Merge similar nodes and edges in the similar business process models to obtain a primary feature model;

[0015] According to the primary feature model, classify all business process models in the business process model library to obtain a model category, and the model category and the business process models belonging to the model category jointly constitute a feature model.

[0016] Further, optimizing the feature model to obtain an optimized feature model comprises:

[0017] According to the frequency of similar nodes and edges in the feature model, layer the feature model and prune low-frequency nodes and edges;

[0018] Convert the pre-obtained domain knowledge structure into nodes and edges and store them in the feature model, and mark the domain label;

[0019] According to the characteristics of the domain to which the feature model belongs, increase the node attributes or edge attributes of the feature model to obtain an optimized feature model.

[0020] Further, obtaining an input model and querying according to the optimized feature model to obtain a query result comprises:

[0021] Match the obtained input model with the optimized feature model to query similar segments of the input model and the optimized feature model;

[0022] When the queried similar segments reach a preset similarity threshold, query in the model category corresponding to the optimized feature model to which the similar segments belong to obtain a query result.

[0023] Further, when matching the obtained input model with the optimized feature model, match from top to bottom according to the hierarchy of the optimized feature model; when the queried similar segments reach a preset similarity threshold, stop matching downward.

[0024] Further, the similar fragment is a process fragment composed of similar nodes and edges of the input model and the optimization feature model.

[0025] Further, the input model is acquired, and the optimization feature model is queried according to the optimization feature model, and the query result is acquired, and the optimization feature model is optimized according to the query result, the input model is classified into the feature model, and the feature model is updated.

[0026] Compared with the prior art, the beneficial effects of the present application are:

[0027] The business process model query method based on the feature model provided by the present application improves the calculation efficiency of real-time calculation by constructing a feature model and preprocessing the business process model in the business process model library. When the feature model is optimized, the matching in the query process can be terminated in advance by layering the feature model, thereby improving the query efficiency. The query accuracy is improved by importing the domain knowledge structure into the feature model. The query accuracy is further improved by increasing the node attributes and edge attributes according to the characteristics of the domain to which the feature model belongs, and considering the similarity of multiple dimensions. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a business process model query method flowchart based on the feature model provided by the present application. DETAILED DESCRIPTION

[0029] The technical solutions of the present application will be further described in detail below with reference to specific implementation manners.

[0030] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as a limitation on the present application. The technical features in the embodiments and the embodiments can be combined with each other without conflict.

[0031] Embodiment one:

[0032] Figure 1 is a business process model query method flowchart based on the feature model provided by the present application. The flowchart only shows the logical order of the method of the present embodiment, and the steps shown or described can be completed in an order different from that shown in other possible embodiments of the present application without conflict. Figure 1

[0033] Referring to Figure 1 , the method of the present embodiment specifically includes the following steps: ​

[0034] Step one: classify all business process models in the business process model library to obtain a feature model;

[0035] Classifying all business process models in the business process model library to obtain a feature model includes the following steps:

[0036] Step A: obtain similar business process models in the business process model library;

[0037] Step B: merge similar nodes and edges in similar business process models to obtain a primary feature model;

[0038] Step C: according to the primary feature model, classify all business process models in the business process model library to obtain a model category, and the model category and the business process models belonging to the model category together constitute a feature model.

[0039] Step two: optimize the feature model to obtain an optimized feature model;

[0040] Optimizing the feature model to obtain an optimized feature model includes the following steps:

[0041] Step a: according to the frequency of similar nodes and edges in the feature model, layer the feature model, and prune low-frequency nodes and edges;

[0042] For similar nodes and edges in the feature model, layer the feature model according to the frequency of similar nodes and edges when they are classified into the feature model, take the nodes and edges with high frequency as the main features of the feature model, and place them in the upper layer of the feature model, take the nodes and edges with low frequency as the secondary features of the feature model, and place them in the lower layer of the feature model. In the process of continuously adding new feature models, the features of the feature model will be updated adaptively, and after accumulating to a higher frequency, in order to limit the size of the feature model, the nodes and edges with low frequency are pruned. When layering the feature model, the number of layers will affect the efficiency of matching using the feature model, therefore, the number of layers can be adjusted using negative feedback for dynamic adaptation.

[0043] Step b: convert the pre-obtained domain knowledge structure into nodes and edges stored in the feature model, and mark the domain label;

[0044] When the text similarity matching of the label fails to achieve the expected effect, for example, the target model cannot be searched, it may be that the professional knowledge related to the near-synonyms in this field is limited, at this time, the domain knowledge structure is used to participate in the matching, and whether the label exists similarity is checked through the relevance of the label on the domain knowledge structure, such as the structural relationship between concepts, the structural relationship between controls, etc., which can all be used as domain knowledge structure. The collected domain knowledge structure is converted into a structure similar to the model, that is, the construction of points and edges, and stored in the feature model. After the feature model is divided into domains, the corresponding domain knowledge structure can be used.

[0045] Step c: according to the characteristics of the domain to which the feature model belongs, adding node attributes or edge attributes of the feature model to obtain an optimized feature model.

[0046] According to the characteristics of the domain to which the feature model belongs, adding node attributes or edge attributes of the feature model can increase the dimension of judging the similarity of the label. For a test process model that tests a certain function point, the test case can be generated from history, which includes a large amount of test input. In software testing, test data input can be divided into types, such as boundary value, equivalence class, etc. The similarity of the label is further considered through the division of these input types. For test data that cannot be divided into data types, the test purpose is explored from the expected result, and the label similarity is considered according to the test purpose similarity. When considering the test process model, the input test data is bound together with the process node. The test data is divided and corresponds to the data design method in software testing, such as equivalence class, boundary value, etc. For data that cannot be classified, the keywords in the expected result are used for classification. This process converts the input test data type into a label, providing a reference for node comparison.

[0047] Step three: obtain an input model and query according to the optimized feature model to obtain a query result;

[0048] Obtaining an input model and querying according to the optimized feature model to obtain a query result includes the following steps:

[0049] Step I: match the obtained input model with the optimized feature model, and query the similar fragments of the input model and the optimized feature model;

[0050] Step II: when the similar fragments queried reach a preset similarity threshold, query the model category corresponding to the optimized feature model to which the similar fragments belong to obtain a query result;

[0051] Step III: according to the query result, the optimized feature model is optimized by negative feedback, and the input model is classified into the feature model, and the feature model is updated.

[0052] The similar fragment is a process fragment composed of similar nodes and edges of the input model and the optimization feature model. When the obtained input model is matched with the optimization feature model, the matching is performed from top to bottom according to the hierarchy of the optimization feature model; when the similar fragments queried reach a preset similarity threshold, the downward matching is stopped. For the similar nodes matched, the adjacent nodes and edges thereof are preferentially matched. The similarity threshold is the proportion of the similar fragments matched in the input model in the optimization feature model, and is used to measure the success rate of finding the similar business process model in the category of the optimization feature model. The similarity threshold is adjusted by using negative feedback. After the business process model query succeeds, the similarity threshold is reduced to terminate the matching with the optimization feature model in advance, thereby improving the query efficiency; after the business process model query fails, the similarity threshold is increased to improve the success rate of the query.

[0053] The business process model query method based on the feature model provided in the embodiment improves the calculation efficiency of real-time calculation by constructing the feature model and preprocessing the business process models in the business process model library. When the feature model is optimized, the matching in the query process can be terminated in advance by layering the feature model, thereby improving the query efficiency; the query accuracy is improved by importing the domain knowledge structure into the feature model; the query accuracy is further improved by increasing the node attributes and edge attributes according to the characteristics of the domain to which the feature model belongs, and considering the multi-dimensionality of the similarity.

[0054] The above is only the preferred embodiment of the present application, and it should be noted that, for those skilled in the art, without departing from the technical principles of the present application, several improvements and modifications can be made, and these improvements and modifications should also be considered as the protection scope of the present application.

Claims

1. A business process model query method based on feature models, characterized in that, include: Classify all business process models in the business process model library and obtain feature models; The feature model is optimized to obtain an optimized feature model; Obtain the input model and perform a query based on the optimized feature model to obtain the query results; All business process models in the business process model library are classified, and the characteristic models obtained include: Obtain similar business process models from the business process model library; Merge similar nodes and edges in similar business process models to obtain a preliminary feature model; Based on the initial feature model, all business process models in the business process model library are classified to obtain model categories. The feature model is composed of the model categories and the business process models belonging to those categories. Optimizing the feature model to obtain an optimized feature model includes: Based on the frequency of similar nodes and edges in the feature model, the feature model is layered, and low-frequency nodes and edges are pruned. The pre-acquired domain knowledge structure is transformed into nodes and edges and stored in the feature model, and then labeled with domain tags; Based on the characteristics of the domain to which the feature model belongs, node attributes or edge attributes of the feature model are added to obtain an optimized feature model; Obtain the input model and perform a query based on the optimized feature model to obtain the query results, including: The obtained input model is matched with the optimized feature model, and similar segments between the input model and the optimized feature model are queried. When the number of similar segments found reaches the preset similarity threshold, the system will search for the model category corresponding to the optimized feature model to which the similar segment belongs, and obtain the search results.

2. The business process model query method based on feature model according to claim 1, characterized in that, When matching the acquired input model with the optimized feature model, the matching is performed from top to bottom according to the hierarchy of the optimized feature model; when the number of similar segments found reaches a preset similarity threshold, the matching stops.

3. The business process model query method based on feature model according to any one of claims 1 or 2, characterized in that, The similar segments are process segments composed of similar nodes and edges between the input model and the optimized feature model.

4. The business process model query method based on feature model according to claim 1, characterized in that, The process of obtaining an input model and querying based on the optimized feature model, and obtaining query results, further includes: performing negative feedback optimization on the optimized feature model based on the query results, classifying the input model into the feature model, and updating the feature model.

Images